Dear X-plor users,
I would like to effectively use Torsion Angle dynamics. I have had
trouble in two respects:
1) I get "too much memory requested" errors, with the result that the
program dies;
2) I have thus far have less-than-spectacular results.
The Details:
With conventional Powell minim and one run of slow-cool, all-atom MD, my
R-Free has dropped to ~28%, 10-1.6 Ang., C222(1), 2 186-residue
monomers/AU, cell constants ~90x95x100 Ang. First strict NCS (R-free
~20%), now NCS restraints.
For 1), if I have just my protein in one chain (186 residues), all is
ok, in that the TAD runs (e.g. at 10000 K) (maxchain was set to 10,
maxtree 1000). If I put in the cofactor, ligand, and 62 water molecules,
xplor dies with the "too much memory requested" error after setting up
the topology (maxchain was set to 70, maxtree 1000). I tried fixing the
waters, and reducing maxchain to 10; in this case xplor ran, but
immediately boosted R-free from ~28% to ~42%, and it did not drop after
that.
Is there any detailed discussion/documentation/explanation about the
various TAD parameters? I have already read Rice & Brunger, and the
various articles about the GroEL refinement. Also, why fix the waters in
TAD? Surely they should be allowed to move as rigid bodies (perhaps with
a harmonic restraint on the Oxygen)?
For 2) In general, on this and two other projects, the R-free doesn't
improve with TAD, even though more conventional Powell or MD does lower
R-free (apparently converging on the correct structure). I have tried up
to 5-10 "trials", at 5000, 10000, and 5000==>300 K. I just don't see the
(expected), better-than-conventional-MD results. What is going on here?
Any help/suggestions would be greatly appreciated.
David Borhani
-----
David Borhani borhani at sri.org
Southern Research Institute Tel: (205) 581-2555
2000 Ninth Avenue South Fax: (205) 581-2726
Birmingham, AL 35205 U.S.A.
